20
Effects of Instructional Support Within Constructivist Learning Environments for Elementary School Students’ Understanding of “Floating and Sinking” Ilonca Hardy Max Planck Institute for Human Development Angela Jonen and Kornelia Mo ¨ller University of Mu ¨nster Elsbeth Stern Max Planck Institute for Human Development In a repeated measures design (pretest, posttest, 1-year follow-up) with 161 3rd-grade students, the authors compared 2 curricula on floating and sinking within constructivist learning environments, varying in instructional support. The 2 curricula differed in the sequencing of content and the teacher’s cognitively structuring statements. At the posttest, both instructed groups showed significant gains on a test on understanding the concepts of density and buoyancy force as compared to a baseline group without instruction. One year later, the group of high instructional support was superior to the group of low instructional support on the reduction of misconceptions and the adoption of scientific explanations. Thus, instructional support within constructivist learning environments fostered elementary schoolchil- dren’s conceptual change in the domain of physics. Keywords: conceptual change, constructivist learning environments, elementary schoolchildren, physics concepts, structural support CONCEPTUAL CHANGE IN SCIENCE LEARNING: THE CASE OF FLOATING AND SINKING It is well documented that many phenomena in the physical world invite children to construct their own explanations sponta- neously (Carey, 2000). However, at the same time, many of the resulting conceptualizations are partially or entirely inadequate from a scientific point of view, constituting what is referred to as misconceptions (for a recent overview, see Chi, 2005). This is, for instance, the case with children’s explanations of why some ob- jects float when immersed into water, while others sink. Interviews conducted by Smith and colleagues (C. Smith, Carey, & Wiser, 1985; C. Smith, Maclin, Grosslight, & Davis, 1997) and Mo ¨ller (1999) with elementary as well as secondary school students showed that children’s explanations often focus on one dimension only: They refer either to the mass of an object (“things that are light will float”), to its volume (“large things will sink”), or to its form (“everything with holes in it will sink”). Moreover, in many explanations, air is seen as an active force that pulls objects upward and water is seen as a force that sucks them downward. These explanations of floating and sinking are incommensurable with scientific explanations (based on the concepts of density and buoyancy force) since, rather than consider the relationship be- tween object and surrounding fluid, the students focus on single properties of objects. Generally, the process of restructuring initial conceptions in order to build conceptions appropriate from a scientific point of view has been framed as conceptual change (e.g., J. Smith, diSessa, & Roschelle, 1993; Vosniadou, Ioannides, Dimitrakopoulou, & Papademetriou, 2001). In this article, we want to contribute to the question of how elementary school students’ conceptual change with regard to floating and sinking may be effectively realized in early science education. We present a study on the long-term effects of two learning environments that allow discovery learning by hands-on experience with authentic material, but which differ in the degree of instructional support. It was investigated whether the sequencing of instructional content and frequency of cognitively structuring statements by the teacher will facilitate students’ conceptual change in the long run. Since scientific explanations of floating and sinking, initiated by the famous principle of Archimedes, were developed by physicists Ilonca Hardy, Center for Educational Research, Max Planck Institute for Human Development, Berlin, Germany; Angela Jonen and Kornelia Mo ¨ller, Institute for Early Science Education, Westfa ¨lische Wilhelms- Universita ¨t Mu ¨nster, Mu ¨nster, Germany; and Elsbeth Stern, Center for Edu- cational Research, Max Planck Institute for Human Development, Berlin. Angela Jonen is now at the Institute for Science Education, Bayerische Julius-Maximilians-Universita ¨t Wu ¨rzburg, Wu ¨rzburg, Germany. This research was supported by the German Research Foundation (DFG) as part of the priority program “The Quality of Schools” (MO 942/1-1; STE 539/9-1). We thank Hella Beister and Rainer Roth for data coding and management as well as Jessica Stehling, Ulrike Austermann, Eva Blumberg, and Daniela Aufderhar for their assistance during the classroom lessons. We also thank Kevin Miller and Fritz Staub for their comments on an earlier version of this article. Finally, we thank the participating teachers and students in Mu ¨nster, Germany, for their enthu- siasm and support of this research. Correspondence concerning this article should be addressed to Ilonca Hardy, Center for Educational Research, Max Planck Institute for Human Development, Lentzeallee 94, 14195 Berlin, Germany. E-mail: [email protected] Journal of Educational Psychology Copyright 2006 by the American Psychological Association 2006, Vol. 98, No. 2, 307–326 0022-0663/06/$12.00 DOI: 10.1037/0022-0663.98.2.307 307

Effects of Instructional Support Within Constructivist Learning … · 2009-03-18 · Effects of Instructional Support Within Constructivist Learning Environments for Elementary School

  • Upload
    others

  • View
    9

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Effects of Instructional Support Within Constructivist Learning … · 2009-03-18 · Effects of Instructional Support Within Constructivist Learning Environments for Elementary School

Effects of Instructional Support Within Constructivist LearningEnvironments for Elementary School Students’ Understanding of

“Floating and Sinking”

Ilonca HardyMax Planck Institute for Human Development

Angela Jonen and Kornelia MollerUniversity of Munster

Elsbeth SternMax Planck Institute for Human Development

In a repeated measures design (pretest, posttest, 1-year follow-up) with 161 3rd-grade students, theauthors compared 2 curricula on floating and sinking within constructivist learning environments,varying in instructional support. The 2 curricula differed in the sequencing of content and the teacher’scognitively structuring statements. At the posttest, both instructed groups showed significant gains on atest on understanding the concepts of density and buoyancy force as compared to a baseline groupwithout instruction. One year later, the group of high instructional support was superior to the group oflow instructional support on the reduction of misconceptions and the adoption of scientific explanations.Thus, instructional support within constructivist learning environments fostered elementary schoolchil-dren’s conceptual change in the domain of physics.

Keywords: conceptual change, constructivist learning environments, elementary schoolchildren, physicsconcepts, structural support

CONCEPTUAL CHANGE IN SCIENCE LEARNING:THE CASE OF FLOATING AND SINKING

It is well documented that many phenomena in the physicalworld invite children to construct their own explanations sponta-neously (Carey, 2000). However, at the same time, many of theresulting conceptualizations are partially or entirely inadequatefrom a scientific point of view, constituting what is referred to asmisconceptions (for a recent overview, see Chi, 2005). This is, forinstance, the case with children’s explanations of why some ob-

jects float when immersed into water, while others sink. Interviewsconducted by Smith and colleagues (C. Smith, Carey, & Wiser,1985; C. Smith, Maclin, Grosslight, & Davis, 1997) and Moller(1999) with elementary as well as secondary school studentsshowed that children’s explanations often focus on one dimensiononly: They refer either to the mass of an object (“things that arelight will float”), to its volume (“large things will sink”), or to itsform (“everything with holes in it will sink”). Moreover, in manyexplanations, air is seen as an active force that pulls objectsupward and water is seen as a force that sucks them downward.These explanations of floating and sinking are incommensurablewith scientific explanations (based on the concepts of density andbuoyancy force) since, rather than consider the relationship be-tween object and surrounding fluid, the students focus on singleproperties of objects. Generally, the process of restructuring initialconceptions in order to build conceptions appropriate from ascientific point of view has been framed as conceptual change(e.g., J. Smith, diSessa, & Roschelle, 1993; Vosniadou, Ioannides,Dimitrakopoulou, & Papademetriou, 2001). In this article, we wantto contribute to the question of how elementary school students’conceptual change with regard to floating and sinking may beeffectively realized in early science education. We present a studyon the long-term effects of two learning environments that allowdiscovery learning by hands-on experience with authentic material,but which differ in the degree of instructional support. It wasinvestigated whether the sequencing of instructional content andfrequency of cognitively structuring statements by the teacher willfacilitate students’ conceptual change in the long run.

Since scientific explanations of floating and sinking, initiated bythe famous principle of Archimedes, were developed by physicists

Ilonca Hardy, Center for Educational Research, Max Planck Institute forHuman Development, Berlin, Germany; Angela Jonen and KorneliaMoller, Institute for Early Science Education, Westfalische Wilhelms-Universitat Munster, Munster, Germany; and Elsbeth Stern, Center for Edu-cational Research, Max Planck Institute for Human Development, Berlin.

Angela Jonen is now at the Institute for Science Education, BayerischeJulius-Maximilians-Universitat Wurzburg, Wurzburg, Germany.

This research was supported by the German Research Foundation(DFG) as part of the priority program “The Quality of Schools” (MO942/1-1; STE 539/9-1). We thank Hella Beister and Rainer Roth for datacoding and management as well as Jessica Stehling, Ulrike Austermann,Eva Blumberg, and Daniela Aufderhar for their assistance during theclassroom lessons. We also thank Kevin Miller and Fritz Staub for theircomments on an earlier version of this article. Finally, we thank theparticipating teachers and students in Munster, Germany, for their enthu-siasm and support of this research.

Correspondence concerning this article should be addressed to IloncaHardy, Center for Educational Research, Max Planck Institute for HumanDevelopment, Lentzeallee 94, 14195 Berlin, Germany. E-mail:[email protected]

Journal of Educational Psychology Copyright 2006 by the American Psychological Association2006, Vol. 98, No. 2, 307–326 0022-0663/06/$12.00 DOI: 10.1037/0022-0663.98.2.307

307

Page 2: Effects of Instructional Support Within Constructivist Learning … · 2009-03-18 · Effects of Instructional Support Within Constructivist Learning Environments for Elementary School

in a process of scientific discovery and justification, studentscannot be expected to discover these principles spontaneously intheir everyday life. Usually, however, students’ participation inconventional science classes is not sufficient for their restructuringof conceptual understanding. Rather, students’ misconceptionshave proven to be quite resistant to change through instruction(Wandersee, Mintzes, & Novak, 1994). Misconceptions have beenconceived of as fragmented, loosely connected pieces of knowl-edge (diSessa, 1988) or as a coherent pattern of notions that formconsistent explanatory systems within domains (e.g., Vosniadou &Brewer, 1992) or even across domains (Chi, 2005). For example,Vosniadou (1994) argues that concepts are embedded in largertheoretical structures constraining them, involving knowledge re-vision at the specific theory level and the framework theory level.Conceptual change thus involves the restructuring of underlyingconcepts, which can take place in various ways depending on theparticular aspect of the concept that is relevant for the respectiveexplanatory context. For instance, an understanding of the conceptof density requires the simultaneous consideration of the twodimensions of mass and volume, where mass needs to be con-ceived of as a continuous and measurable characteristic of thematerial world—a conception that young children often lack (C.Smith et al., 1985). In order to apply fully the concept of densityin the context of floating and sinking of objects, a further concep-tual shift needs to occur. Rather than its attribution to the charac-teristics of single objects, the relationship between object andsurrounding fluid needs to be considered causal for objects’ float-ing and sinking. Both the comparison of densities (of object andfluid) and the comparison of forces (of gravity and buoyancy) thusrequire the simultaneous consideration and integration of concepts.

Frequently, the topics of density and buoyancy force are intro-duced only in secondary school, based on the argument thatstudents need to be able to grasp the formal aspects of the involvedformulas such as proportions. Although particularly young chil-dren tend to adopt an undifferentiated concept of weight anddensity (C. Smith et al., 1985), even some secondary schoolstudents approach instruction with this type of commonsense the-ory (C. Smith et al., 1997). A study by Kohn (1993) showedsimilarities even between preschoolers and adults with regard totheir inadequate strategies for judging an object’s floating orsinking. Nevertheless, there is evidence that conceptual advance-ment may be reached by curricula addressing these conceptsexplicitly. Most of the curricula developed for early secondaryschool focus on conceptual change with regard to a differentiationof students’ conceptualization of matter. For example, C. Smith etal. (1997) showed that employing matrices of squares with dots toillustrate the densities of different materials during an instructionalunit helped many students learn to differentiate between weightand density and achieve an integrated understanding of density(see also C. Smith, Snir, & Grosslight, 1992). Lehrer, Schauble,Strom, and Pliggie (2001) had fifth graders model material kind inlinear graphs after introducing them to the mathematical conceptsof volume measure and similarity. Generally, the judgment ofvolume as a prerequisite to understand the relation of volume andmass does not seem to present difficulties to students of elemen-tary school age (Halford, Brown, & Thompson, 1986); rather,within a curriculum on floating and sinking, students need to begiven opportunities to discover that the amount of water displacedis dependent on an object’s volume and not its mass. This under-

standing of water displacement may then be further differentiated,allowing for the comparison of densities and the determination ofbuoyancy. In our curriculum, we therefore focus on a basic un-derstanding of water displacement, which then can support theconstruction of more complex scientific explanations of floatingand sinking. Commonly, children are introduced in elementaryschool to a basic concept of material kind, that is, the realizationthat solid objects of the same material behave the same way whenimmersed in water. If, however, children were also introduced tothe explanations for the behavior of different materials in water,thus receiving the opportunity to revise misconceptions early on,there is good reason to expect that they will be able to profit morefrom the formulas of density and buoyancy force treated in sec-ondary school.

CONCEPTUAL CHANGE WITHIN ACONSTRUCTIVIST PERSPECTIVE

To be able to successfully restructure their concepts, studentsneed to engage in a process of knowledge integration (Davis, 2003;Linn, 1995); that is, they need to link newly constructed scientificideas to their current concepts, refining or dismissing them in thecase of incommensurability. Conceptual change research hasshown that this process of integration is by no means a sudden shiftfrom a naive explanation to a scientifically adequate view butrather a gradual process of restructuring interrelated concepts, inthe course of which misconceptions may reoccur depending on theparticular context (e.g., Caravita, 2001; Tyson, Venville, Harrison,& Treagust, 1997; Vosniadou et al., 2001). Why and under whatconditions does successful conceptual change occur? According tothe pioneering work of Posner, Strike, Hewson, and Gertzog(1982), an essential condition of conceptual change is for learnersto become dissatisfied with the conceptions they have, thus expe-riencing the need for new explanatory mechanisms. Central to thisnotion is the induction of cognitive conflict as a motor of concep-tual change, starting with the initial conceptions a child may hold(Chinn & Brewer, 1993). For instance, in order to challenge themisconception that all floating objects contain air, one may con-front children with a hollow, yet sealed object that sinks. Suchkinds of confrontations are crucial for challenging plausible butinappropriate explanations, and as a consequence, students maybecome receptive to new ideas.

There is agreement among psychologists and educational scien-tists that the adoption of appropriate scientific explanations is aconstructive process requiring the active cognitive engagement ofindividuals rather than a process of direct transmission. As aconsequence, learning environments in science and mathematicsthat honor constructivist processes of knowledge acquisition haveat their core the creation of opportunities for students’ activesensemaking, allowing them to discover new principles and ex-planations of the topic at hand (Bransford, Brown, & Cocking,1999; Cobb, 1994; Greeno et al., 1998). In a typical learningenvironment of this kind, a group of learners is confronted withcomplex, challenging, and authentic tasks and materials that en-courage activities such as experimentation, exploration, and groupgeneration of hypotheses and explanations. It is expected thatstudents are able to self-regulate their learning to a degree thatallows them to make sense of the material and physical resourcesaccording to their individual or common prior knowledge and

308 HARDY, JONEN, MOLLER, AND STERN

Page 3: Effects of Instructional Support Within Constructivist Learning … · 2009-03-18 · Effects of Instructional Support Within Constructivist Learning Environments for Elementary School

capabilities. The social negotiation and common work on a task isthought of as a vehicle for the sharing of opinions, observations,and preexisting beliefs, adding to the anchoring of new knowledge.

Several studies suggest that students participating in a learningenvironment constructed according to the principles of construc-tivist learning achieve a higher degree of conceptual understandingin science and mathematics than students in environments of directinstruction (Christianson & Fisher, 1999; Staub & Stern, 2002;Tynjala, 1999). However, many open questions remain as to howconstructivist ideas of learning can be effectively realized atschool. In an insightful, recent article, Mayer (2004) points out thatthe translation of principles of constructivist learning into learningenvironments has largely followed the simple formula “construc-tivism � hands-on activity” (p. 17). As a consequence, it mayhappen that, during an open-ended and thus highly self-directedlearning activity, students do not actually reflect on the relevantconcepts, even if the material provided is closely relevant to thephenomenon to be studied. If a setting is too complex and the tasksleave students too much freedom as to their selection and progres-sion through them, students may be actively involved; neverthe-less, they cannot arrive at the scientific conclusions intendedbecause their sensemaking is not integrated into a conceptualframework. It is also at the point of addressing misconceptions thatthe question arises whether purely activity-oriented learning envi-ronments will be cognitively challenging enough for all students toenable them to proceed to the validation, correction, or finalrejection of these deeply held misconceptions. Under the consid-erations outlined above, it seems likely that discovery learningenvironments can contribute to conceptual change; crucially, how-ever, learning environments based on constructivist principlesneed to be designed in a way that supports students’ cognitiveactivity apart from behavioral activity. That is, behavioral activitymay be a vehicle for supporting cognitive activity, but it need notnecessarily translate into the cognitive activity necessary for thesuccessful conceptual integration of insights into prior understand-ing. An important issue for research, therefore, concerns the ques-tion of how aspects of learning environments may be successfullystructured. Moreover, teachers willing to apply principles of con-structivist learning in their classrooms must achieve more clarityabout the conditions in which their feedback and prompts can helpstudents to further conceptual understanding. We intend to shedlight on these questions.

THE ROLE OF INSTRUCTIONAL SUPPORT INCONSTRUCTIVIST LEARNING ENVIRONMENTS

Which are the dimensions relevant to instructional supportwithin constructivist learning environments? Recently, the con-struct of scaffolding, originally framed by Wood, Bruner, and Ross(1976) and Vygotsky (1978), has been reexamined in its definitionwithin complex science learning environments (Davis & Miyake,2004; Hogan & Pressley, 1997). Reiser (2004) proposed the twomechanisms of structuring and problematizing as essential for thescaffolding of science tasks: While structuring works to reduce thecomplexity of a task, for example, by decomposition of a complextask and reduction of the degrees of freedom, problematizingsubject matter works to relate students’ work to a disciplinaryframework, for example, by eliciting elaborations or by addressingdiscrepancies and disagreements in science discourse. Pea (2004)

refers to the processes of channeling and focusing students’ atten-tion on relevant aspects of a task on the one hand, and to modelingmore advanced solutions on the other hand, as critical aspects ofscaffolding. The two key elements of instructional support, framedas scaffolding, thus seem to be (1) the structuring of tasks to allowstudents to remain focused on important aspects and (2) the sup-port of students’ reflection on their insights within a larger contextof scientific reasoning.

Teacher prompts may be one way in which students can besupported to reflect on their scientific understanding. For example,Davis and Linn (2000) found that using prompts encouragingreflection significantly increased eighth graders’ conceptual un-derstanding of thermodynamics and light in a computer-supportedlearning environment (see also Davis, 2003). Scaffolding within anentire classroom may involve the teacher providing refocusingstatements, challenges to assumptions or interpretations made bystudents, or advance organizers directing students’ attention tocertain aspects of a phenomenon to be investigated subsequently(Hogan & Pressley, 1997).

In sum, instructional support should help students to move alonga sequence of conceptual development during which they canactively discover possible inadequacies of initial conceptions aswell as convincing new explanations. As Reiser (2004) points out,it is a delicate task to create an optimum level of challenge throughthe provided support so that learners continue to engage actively inthe learning process. In order to create this level of support, ateacher needs to be tuned to the needs of the learners in advance,anticipating their learning trajectories in the appropriate design oflearning tasks. However, despite the need for individually tunedsupport of student reasoning processes, scaffolding environmentscan be achieved for an entire class if the social processes ofprompting and modeling scientific reasoning support and reflecton individual sensemaking of the topic at hand (Hogan & Pressley,1997).

THE CURRENT STUDY: EXPERIMENTALVARIATION OF INSTRUCTIONAL SUPPORT ON

THE TOPIC OF “FLOATING AND SINKING”

In our study, we investigated the effects of different degrees ofinstructional support within constructivist learning environmentswith respect to elementary school students’ conceptual understand-ing of floating and sinking. We employed a repeated-measuresdesign (pretest, posttest, and 1-year follow-up measure of concep-tual understanding of floating and sinking), with two groups basedon an experimental variation of instructional support (three third-grade classrooms each) and one baseline group which received thetests only (two third-grade classrooms). Our operationalization ofinstructional support was based on two central aspects of scaffold-ing, supporting (1) students’ focusing on relevant aspects of thetasks and (2) students’ reflection on their insights and scientificreasoning processes. It needs to be pointed out again that bothlearning environments designed in this study were based on con-structivist principles of learning, with a high degree of students’active experimentation with relevant material in both experimentalgroups, allowing them to discover relationships between physicalquantities and scientific principles by independent group work andto discuss hypotheses and discoveries in sessions with the entireclass. Between the two experimental groups of high instructional

309EFFECTS OF INSTRUCTIONAL SUPPORT

Page 4: Effects of Instructional Support Within Constructivist Learning … · 2009-03-18 · Effects of Instructional Support Within Constructivist Learning Environments for Elementary School

support (HIS) versus low instructional support (LIS), the instruc-tional time, the teacher, and the material for experimentation werenot varied.

Variation of instructional support with regard to the task envi-ronment concerned the sequencing of instructional content intounits which followed an anticipated path of conceptual develop-ment with regard to floating and sinking. While in both learningenvironments the investigation of the question of “Why does alarge ship of iron float in water?” introduced the topic of floatingand sinking, this topic was segmented into smaller units only in thegroup of HIS. That is, a complex learning environment allowingfor experimentation with regard to different aspects of floating andsinking—such as material kind, density, water displacement, orbuoyancy force on different learning sites—was used in bothgroups; however, it was prestructured into a sequence of experi-mental activities moving from more basic concepts such as mate-rial kind to the integrated concepts of the comparison of densitiesand relation of buoyancy force and gravity in the group of HIS. AsVosniadou et al. (2001) suggest, a central aspect of environmentsconducive to conceptual change is the consideration of the order inwhich concepts are acquired, based on the logical structuring of theinvolved domain. In the case of floating and sinking, the conceptof material kind as an explanation of everyday life is fundamentalto understanding more advanced concepts such as the comparisonof densities. Moreover, many initial misconceptions may be re-jected on the basis of evidence concerning the behavior of objectsof the same material but different mass and volume in water. Wetherefore included lessons allowing these types of conclusionsearly on in the group of HIS, then advanced to the scientificconcepts of density, water displacement, and buoyancy force, withan attempt to integrate these concepts in the last lesson. The groupof LIS, in contrast, was free to use any experimental settingextensively within the context of researching the question aboutthe iron ship. Here, students could work on activities with variousfoci such as density or water displacement, researching a certainself-generated issue of investigation; it is important to note, how-ever, a complex question was not segmented in advance intosmaller conceptual units.

With regard to the support of students’ reflective processes,discussions involving the entire class were scaffolded to a higherdegree by the teacher in the group of HIS. That is, the teacherintended to activate students cognitively with statements relatingand contrasting ideas or hypotheses proposed by different students,by once more referring to earlier misconceptions, or by introducinga hypothesis or observation that the students themselves had notconsidered. In contrast, in the group of LIS, group discussionswere much more student centered, with students themselves react-ing to each other’s statements and challenging each other’s hy-potheses. The role of the teacher was closer to that of organiza-tional supervision, with a lower frequency of content-relatedprompts to students’ reasoning processes.

HYPOTHESES

We hypothesized that through participation in a learning envi-ronment based on constructivist learning theories, third graderswill show achievement growth in their understanding of floatingand sinking that can be characterized as conceptual change. Thismeans that, after the instructional unit, learners will show evidence

of having rejected misconceptions and adopted explanations com-patible with the scientific concepts of density and buoyancy force.Students are expected to make decisions about the floating andsinking of objects based on their differentiated conceptualizationswith regard to density as a feature of substances, allowing thecomparison of densities, and with regard to buoyancy force asresulting in a relational system of object and displaced water.

Due to the test-taking experience and informal exposure to thetopic of floating and sinking, we expected some gains in theadoption of so-called explanations of everyday life, which arebased on concepts of material kind and hollowness, even in thebaseline group. However, neither those students’ misconceptionsnor their scientific explanations were expected to change signifi-cantly in comparison to students who attended the learningenvironments.

Our major hypothesis concerned the difference between the twogroups of instructional support. We expected the group of HIS tooutperform the group of LIS in both aspects of conceptual change,that is, by the decrease of misconceptions as well as by an increaseof appropriate scientific explanations. Moreover, we expected thesuperiority of the group of HIS to be reflected by an integratedconceptual understanding at the individual level, where studentswill simultaneously hold misconceptions and appropriate concep-tions to a lower extent than in the group of LIS. As a consequenceof a more integrated understanding in the group of HIS, weexpected long-term effects in the 1-year follow-up measure for thegroup of HIS, indicating fundamental conceptual change with alower degree of reemergence of misconceptions.

METHOD

Design and Procedure

In a 3 (time) � 3 (group) repeated measures design, we investigated theeffects of instructional support (group of HIS, group of LIS, baseline groupwith repeated test administration only) on third graders’ conceptual under-standing of floating and sinking. The intervention was based on twoeight-lesson curricula which were preceded by a pretest and followed by aposttest one week after the end of instruction as well as a follow-up test oneyear after the instruction had taken place. In addition to the repeated Teston Floating and Sinking, we employed a transfer test on the application ofprinciples of floating and sinking within three days after the administrationof the posttest. At the posttest, the teachers of all classes were instructednot to revisit the topic of floating and sinking during the upcoming schoolyear. One teacher of a class of LIS reported to have revisited the topicshortly after the posttest in the course of a class excursion on a ferry. Thestudents were not told that there would be a 1-year follow-up test; only theteachers were informed shortly before the date.

In order to ensure the implementation of our instructional variation, allof the instruction was realized by Angela Jonen, who was educated as anelementary school teacher. Angela Jonen had not taught in any of the threeschools before. The entire instructional unit was taught in 2-week intervals,alternating classes with HIS and LIS in one school at a time. At eachschool, the two classes were randomly assigned to either the condition ofHIS or LIS. All of the instruction was videotaped.

Participants

A total of 161 third graders in eight intact classrooms participated in thisstudy. The original sample size of 185 was reduced only by students whomissed one of the tests or important parts of the curriculum in third grade,or else were not any more part of the same class in fourth grade at the

310 HARDY, JONEN, MOLLER, AND STERN

Page 5: Effects of Instructional Support Within Constructivist Learning … · 2009-03-18 · Effects of Instructional Support Within Constructivist Learning Environments for Elementary School

1-year follow-up. All classes came from elementary schools located inmiddle-class neighborhoods of a midsize town in Germany. The total ofeight participating classes (six experimental classrooms, two baselineclassrooms) were selected because they were composed of students withmixed social background (according to the school district), they showed ahigh variance in students’ cognitive capabilities, and students were ex-pected to have experienced discovery learning as well as traditional tech-niques in their instruction. That is, students were used to some forms ofopen, hands-on experimentation in their instruction but had not been taughtexclusively by this method. We observed a total of three science lessonsinstructed by the respective classroom teachers before the beginning of thestudy to ensure that the type of instruction that students were used to wasindeed comparable. The observational instrument focused on the teacher’semployment of a variety of organizational forms (such as group work,experimental learning sites, and traditional instruction) and the treatment ofcomplex science topics that students could investigate in a self-directedway.

The group of HIS comprised 66 students (26 girls, 40 boys; mean age inthird grade of 9.02 years) from three classes, and the group of LIScomprised 59 students (28 girls, 31 boys; mean age in third grade of 8.95years) from three classes. The baseline group was composed of 36 students(23 girls, 13 boys; mean age in third grade of 9.33 years) from twoclassrooms of the same school.

Curriculum

We developed two instructional units on the topic “Why does a largeship of iron float?” which were both based on constructivist principles oflearning but varied in the degree of instructional support provided tostudents. Between the two instructional units we did not vary the material(Styrofoam, wood, metal, wax, modeling clay, etc.) that was available tostudents either as chunks of varying size and form or as everyday objects(metal plates, pins, wooden buttons, candles, nails, etc.) in order to enabletheir investigation of the topic of floating and sinking. Apart from everydayobjects, the material for experiments included specifically designed objectssuch as cubes of the same volume but different material, objects of thesame mass but different volume, and hollow and solid objects of the samemass and material. These objects were constructed to confront typicallyheld misconceptions such as the concentration on the mass of objects. Inthe two instructional environments, we also employed the same experi-mental learning sites, which consisted of experiments on specific aspectsrelated to floating and sinking such as water displacement. These sites wereprepared in advance and were designed as independent student hands-onactivities that addressed specific research questions, including directionsfor students’ handling of material and hints to draw students’ attention tospecific features and outcomes of their investigations. The experimentallearning sites concerned the topics of material kind, density, water dis-placement, and pressure or buoyancy force. Both instructional units com-prised eight 90-min lessons. The fifth lesson in the sequence was the samein both instructional units; this lesson concerned the topic of water pressureand took place in a public swimming pool in order to enable students toexperience water pressure on large objects and on their own bodies.

In the instructional unit of LIS, all of the objects and chunks of differentmaterial as well as the experimental learning sites were available tostudents throughout the entire instructional unit (except for the fifth lessonin the swimming pool). Students could use the material as they wanted toinvestigate their own research questions, which they derived from answer-ing the overarching question of why a large ship of iron floats in water, ina self-directed way. They could also work on the experimental learningsites, which were designed for students to get ideas about how to adapttheir own experiments, to draw attention to phenomena not yet investigatedor perceived by the students, and to instigate discussion between studentson explanatory mechanisms for floating and sinking. During students’experimental work, the teacher gave individual support to students con-

cerning the process of investigation. She also provided written feedback onstudents’ learning diaries in which they recorded their investigations,explanations, and further questions. It was up to the students to modifytheir research agenda depending on the insights they had gained duringexperimental work and group discussions. Students presented results oftheir investigations in small groups and in discussions with the entire class.The discussions were student centered; that is, students themselves wouldintroduce new explanations, hypotheses, and problems encountered duringtheir research, add to each other’s interpretations, and challenge eachother’s hypotheses. The teacher enabled the process of discussion organi-zationally but only rarely contributed to discussions in terms of the inves-tigated content.

In the instructional unit of HIS, the availability of objects of differentmaterial and experimental learning sites was determined by the specificsubtopic to be investigated in the particular lesson. In this way, thecomplex research question of why a large ship of iron floats was parti-tioned into subordinate research questions, all of which were assumed tobuild toward an integrated understanding of floating and sinking. Lessonsproceeded from the concepts of material kind; the differentiation of lightand heavy material (or density), water displacement, and buoyancy force tothe integration of these aspects in the relation between buoyancy force andgravity and the comparison between densities. Table 1 summarizes thesequencing of content in the HIS environment. While students worked onthe experimental learning sites, the teacher would provide individual sup-port as needed. During discussions with the entire class, the teacherfrequently structured interaction on a content level, that is, attempted tocognitively activate students in making connections to their own knowl-edge base. For example, the teacher highlighted contradictory statements;asked for causal explanations, validations, or applications; or gave sum-maries of results. Table 2 summarizes the central characteristics of the LISand HIS environments with respect to variation in students’ work with theexperimental learning sites and class discussions. In addition, the Appendixcontains the coding scheme used to rate teacher and student discourseduring class discussions, which served as an implementation check fordifferences in the two learning environments. This scheme provides insightinto the exchanges that took place, including examples of teacher andstudent comments. We expected a higher frequency of teacher structuringstatements in the group of HIS and a higher frequency of student contri-butions to other students’ statements in the group of LIS.

Pilot Study

We conducted pilot studies in a total of four classrooms in order tooptimize the implementation of the two curricula. Since we opted for thesame experienced elementary school teacher, Angela Jonen, to manageboth instructional variations, we needed to ensure that she could practicethe variation of her teaching style. Her style of teaching is grounded in aconstructivist view of learning, where she tends to hold back herself asmuch as possible during learning activities and group discussions in orderto enable students’ active construction of meaning. She thus tended to astyle of teaching more in line with the curriculum of LIS, and she neededto get experience especially in the degree to which she was to providecognitively structuring statements in the group of HIS (e.g., in terms ofprompting for misconceptions or contrasting students’ answers in groupdiscussions). Optimization of the two curricula further concerned theoptimal arrangement and wording of tasks for the experimental activitiesperformed in both instructional groups. That is, since the instructionalvariation consisted only of the additional sequencing of tasks in the groupof HIS, it was necessary to ensure for both curricula that students couldwork with the experimental learning sites on their own. Observationsconducted by research assistants and one of the primary researchers in thepilot classrooms confirmed the differing implementations according to thedegree of instructional support.

311EFFECTS OF INSTRUCTIONAL SUPPORT

Page 6: Effects of Instructional Support Within Constructivist Learning … · 2009-03-18 · Effects of Instructional Support Within Constructivist Learning Environments for Elementary School

Instruments

In order to address our hypothesis on the effects of instructional support,four instruments were administered. First, and most important, the Test onFloating and Sinking was administered to students as the pre-, post-, andfollow-up test. Second, in order to obtain evidence of the construct validityof the Test on Floating and Sinking as a measure of students’ conceptualunderstanding of density and buoyancy force, we administered a transfertest on floating and sinking shortly after the posttest. Third, in order toensure that the instructional variation of HIS and LIS was actually imple-mented in the intended way, videotaped lessons were rated by trainedpersons not informed about the experimental variation with regard toteacher activities (implementation check). Fourth, the instructional dis-course in each class was categorized according to criteria such as struc-turing statements by the teacher (implementation check).

Test on Floating and Sinking

Test Content

A test of students’ explanations with regard to floating and sinking, com-prised 36 items, was employed as pretest, posttest, and follow-up test. Three

items were free-response items, while 33 were multiple-choice items thatvaried in the number of answers to be confirmed. Seventeen items weretrue-false items; that is, students were asked to judge the correctness of a singlestatement by checking it. Sixteen items were multiple-choice items with fourto seven possible answers, and the participants were asked to confirm thosethey considered to be true. The test was developed on the basis of elementaryschool students’ spontaneous explanations derived from interview data, in-cluding formulations regarding the concepts of material kind, density, andbuoyancy force in child-appropriate language, which were adapted in severalcycles of pilot testing. The use of child-appropriate language also means thatwe did not employ the unfamiliar terms of mass and volume; instead, we usedthe colloquial terms of weight and size for these concepts. In addition, themeaning of the new term of water displacement was introduced briefly beforethe test. The ordering of items in the test was determined by content consid-erations. Free-response items were presented in the beginning of the test toallow for maximum variation of student language; multiple-choice items werepresented in a mixed order to ensure variation of item contexts. Additionally,the item formats requiring specific instructions were placed in the first half ofthe test so the teacher could address the entire class during explanations. Theitems were constructed to test two broad areas of understanding.

Table 1Sequencing of Content in the High Instructional Support Environment

Lesson Topic Learning sites and activities

1 “Why does a large ship of iron float?” Form initial hypotheses; work with solid objects; orderobjects according to floating and sinking in water

2 “What floats, what sinks?” (material kind) Work with solid objects, address misconceptions;order objects according to material

3 “Why does wax float but metal sink?” (density) Work with objects of same volume and different mass;differentiate mass and volume

4 “What happens to water when an object is immersed into it?”(water displacement)

Work with solid and hollow objects, varying involume and mass

5 Excursion to the swimming pool: “What does the water do toobjects?” (buoyancy force)

Work with large hollow objects and own body,experiencing buoyancy force

6 “What does the water do to objects?” (water displacementand buoyancy force)

Work with solid and hollow objects on buoyancy forceand relation to water displacement

7 Relation of water displacement and buoyancy force Work with modeling clay (modeling of boats) on therelation of water displacement and buoyancy force

8 Relation of buoyancy force and gravity: “Why does a largeship of iron float?”

Represent forces by children pulling on each other’shands; work with large hollow objects

Table 2List of Characteristics of High Instructional Support and Low Instructional Support Environments

Characteristic High instructional support Low instructional support

Experimental learning sitesLearning material Differently shaped and sized material and everyday objects Differently shaped and sized material and everyday

objectsContent Sequenced into eight consecutive units; availability only of

learning sites concerning one subtopicAvailability of all experimental learning sites and

possibility to design own experimentsTeacher Managerial and contentwise structuring statements Managerial and contentwise structuring statementsStudents Document their insights on handouts providing research

questions and structuring comments for each learningsite

Document their experiments and insights in alearning diary, using provided handouts asneeded

Social forms of learning Individual or group experimentation at learning sites withtime limit for each learning site

Individual or group experimentation at learning siteswith no time limit within the lesson; spontaneousgroup discussions

Class discussionsStudents Mainly react to teacher statements; free to react to other

studentsFree to initiate topics; need to call on other students

Teacher Teacher-directed with contentwise structuring comments;topic is focused on content of each lesson

Teacher keeps conversation going

Time About 50% of entire unit About 50% of entire unit

312 HARDY, JONEN, MOLLER, AND STERN

Page 7: Effects of Instructional Support Within Constructivist Learning … · 2009-03-18 · Effects of Instructional Support Within Constructivist Learning Environments for Elementary School

1) Water displacement. A group of multiple-choice items (N � 9),separated into three contexts, focused on the topic of water displacement asa core concept from which other explanations of floating and sinking canbe derived. In these items, the misconception of dependency of amount ofwater displacement on mass, not volume, is tested. Figure 1 shows asample item confronting students with information about an object’s massand volume to determine the water displaced by it.

The three items of the second context showed four glasses, each ofwhich contained a ball of the same volume but different mass than theothers. It required students to draw lines for the respective water levels,given the water level for one ball in the first glass. The three items of thethird context questioned students’ judgment about which of two balls ofdifferent volume and mass would displace more water, with the mass of theobjects displayed qualitatively by a balance beam.

2) Explanations of floating and sinking. The second group of itemsfocused on students’ explanations of floating and sinking. This group wasfurther divided into two areas.

2a) Explanations of specific situations. This group of items focused onstudents’ explanations of specific objects’ behavior when immersed intowater. The objects, which were always shown to students and depicted inthe test, were chosen to be extreme cases with regard to volume, mass, orform. For each item, it was made clear that objects were first entirelyimmersed into water before releasing them. This group included sevenmultiple-choice items and the three free-response items. The multiple-choice items incorporate typical misconceptions of elementary schoolstudents (such as reference to an object’s mass, volume, form, or air as anactive force) as well as explanations derived from everyday life (such asreference to an object’s material or its hollowness, or mentioning of therole of water in floating) and scientific concepts (such as water displace-ment, density comparisons, and buoyancy force). In order to avoid con-fronting students with unfamiliar vocabulary, all explanations were testedcarefully to be understandable to elementary school students, yet at thesame time scientifically appropriate in the relationships expressed. Wordssuch as lighter/heavier than or pushes up, which were used repeatedly ininstruction, were offered in a correct and an incorrect version within oneitem. For each of the situations, the students first had to decide whether theobject would sink or float in water; then they had to check the explana-tion(s) judged as most appropriate. The seven situations to be judged bystudents concerned the floating or sinking of a wooden button, a flat pieceof metal, a flat piece of Styrofoam, a piece of metal wire, a pin, a woodenblock, and an iron ship.

The two items shown in Figure 2 exemplify the construction of themultiple-choice items. The item referring to the wooden button includesprominent misconceptions regarding the mass and form of the object(Answers 2 and 3), an explanation of everyday life (Answer 5), and thescientifically correct explanations with the concepts of density (Answer 6)and buoyancy force (Answer 1), along with one worded in the oppositedirection (Answer 4). The scientific explanation based on a comparison ofthe object’s mass with its displaced water (Answer 6) refers to the totalamount of water that may be displaced by an object due to its volume, thusallowing a comparison of mean densities of object and water. Therefore, anobject floats if the amount of water that can be displaced weighs more thanthe object itself, or the density of the water is greater than the density of theobject. In contrast, the actual amount of water that is displaced by a floatingobject weighs the same as the object, a relationship referred to as theprinciple of Archimedes. Students confronted with this item first need tomake a choice about the object’s behavior in water and then need to checkthe explanations they find most convincing. Since the conceptual changeliterature shows that the integration of new explanations into existingframeworks is a continuous process, the possibility of multiple answerswas given to students. The item referring to the piece of metal wire isconstructed according to similar principles. Two scientific explanationsrefer to density, one worded in a correct way (Answer 5) and one wordedin the opposite direction and therefore incorrect (Answer 3).

The three free-response items referred to questions explaining whathappens if (1) a large tree trunk, (2) a wooden board with holes, and (3) aship of iron are immersed into water. Again, students had the opportunityof giving multiple explanations for the objects’ floating or sinking in water.

2b) Generalized concepts. A third group of items consisted of a totalof 17 true-false statements with respect to explanatory mechanisms offloating and sinking or principles generalized beyond specific objects. Forexample, students were asked to check whether statements such as “Heavyobjects cannot float in water,” “Things without air cannot float in water,”“Some types of material always float in water,” “Sinking objects are beingsucked down by the water,” “Water pushes more against large than smallobjects,” or “Any hollow object will float in water” are correct. This groupof items also included five items testing students’ generalized understand-ing of the concept of material kind, asking them to make predictions aboutfloating or sinking of a certain material, given information about its size(see Figure 3).

Figure 1. Sample item of the content area of water displacement.

313EFFECTS OF INSTRUCTIONAL SUPPORT

Page 8: Effects of Instructional Support Within Constructivist Learning … · 2009-03-18 · Effects of Instructional Support Within Constructivist Learning Environments for Elementary School

Test AdministrationAll tests were administered by Angela Jonen in the respective class-

rooms. All students of a class took the tests at the same time. The studentswere presented with a test booklet. The test administrator read out loudsample items of different test formats (multiple-choice items, true-falseitems, and free-response items) and explained how to proceed in answeringthem. The administrator demonstrated the material or the situation de-scribed in the sample items, with all of the objects described in the testlocated at the front desk. Students were free to look at and pick up all theitems described in the test throughout test taking as they wished. In thepretest, the term water displacement was explained to students by showinga cube, immersing it into water, and asking whether there will still be water

in the place where the cube was now. Then, the term displaced water wasintroduced to specify the amount of water which was replaced by the cube.Thus, no explanations of the physical principles underlying water displacementwere given. Also, the teacher demonstrated how an object was entirely im-mersed in water in order to clarify the situations in which students shoulddecide on an objects’ floating or sinking when immersed in water. There wereno demonstrations of objects’ actual floating or sinking, nor were the studentsallowed to try out objects’ behavior in water. Students were free to work on theitems at their own pace. Each student took the test with the same order ofitems. Altogether, it took approximately 60 minutes to administer the entiretest. Although the students took the test very seriously, its administration tookplace in a relaxed atmosphere.

Figure 2. Sample item of the content area of explanations of specific situations of floating and sinking.

Figure 3. Sample item on the concept of material kind.

314 HARDY, JONEN, MOLLER, AND STERN

Page 9: Effects of Instructional Support Within Constructivist Learning … · 2009-03-18 · Effects of Instructional Support Within Constructivist Learning Environments for Elementary School

Coding scheme for free-response items. We scored answers on free-response items on three broad levels of conceptual understanding, with thehighest level separated into two sublevels. Each student answer could bescored on multiple levels.

1) Misconceptions. This level of explanation was scored if studentsanswered by a one-dimensional focus on, for example, mass, volume, orshape, or if they considered air an active force. Also, students whoindicated with a question mark that they did not have an explanation wereincluded in this category. Examples of student answers to the question ofwhat happens if a wooden board with holes is immersed into water include“sinks because it has holes and it is really flat,” “sinks because the waterwill come through the holes and lies on the board and then the board willbe heavier,” and “floats because the weight of the board is spread out.”

2) Explanations of everyday life. This level of explanation includesanswers mentioning the role of the water (as the object being lighter orheavier than water), the concept of material kind, and the hollowness ofobjects as explanations of floating and sinking. Examples of studentanswers are “floats because it is wooden,” “floats because water weighsmore,” and “floats because this board with holes is made of wood. And itdoesn’t matter whether the wooden board is with or without holes.”

3a) Scientific explanations based on one concept (Scientific Explana-tions 1). Answers that consider one of the physical quantities of density,water pressure, or buoyancy force were scored on this level of conceptualunderstanding. Examples of student answers include “floats because thewater wants back its place” and “floats because the water pushes the boardup.”

3b) Scientific explanations based on more than one concept (ScientificExplanations 2). Answers that consider two of the physical quantities ofdensity, water pressure, or buoyancy force simultaneously, or in which allthree aspects are integrated into one explanation, were scored on thishighest level of explanation. Examples of student answers are “floatsbecause the holes don’t matter and because the displaced water weighsmore than the wooden board,” “floats because the displaced water weighsmore than the board, it is pushed up,” “floats because the board displacesas much water as it weighs,” and “floats because it is lighter than the sameamount of water.”

While Levels 1, 2, and 3b correspond to the explanations offered bymultiple-choice items, there is no level corresponding to Level 3a inmultiple-choice items. Since this lower level of scientific explanationsturned out to be a frequent answering level of students, we decided to splitthe level of scientific explanations into the two sublevels 3a and 3b.Interrater reliabilities on each of the four levels of rating were � .95, withtwo trained persons each rating all of the free-response items.

Scoring

We scored the test with a total score indicating students’ overall answer-ing patterns on the test and separate scores for different levels of concep-tual understanding of floating and sinking and water displacement. Thesescores were computed separately for multiple-choice and free-responseitems.

1) Separate levels of conceptual understanding. With the separatescores of conceptual understanding, we can evaluate to what extent stu-dents gave up misconceptions and adopted everyday life or scientificexplanations independently of each other. Since conceptual understandingat each level may vary depending on the selection or production ofrespective answers, we computed separate scores for multiple-choice (MC)and free-response (FR) items. This decision is justified by the low tomedium correlations between ratings of free-response items with respec-tive conceptual levels of multiple-choice items, ranging from .13 to .38 inthe pretest, from .30 to .42 in the posttest, and from .25 to .39 in thefollow-up test. Apart from two pretest correlations, all correlations aresignificant ( p � .001).

For the content area of “water displacement,” we computed a sum scoreof correct answers (Water Displacement, MC), with a maximum of 9 points

and Cronbach’s alpha (pretest, posttest, follow-up test) � .79, .89, .88,respectively. Here, we did not differentiate between different levels ofconceptual understanding, since a correct answer on these multiple-choicequestions means that a student correctly considered volume instead of mass(incorrect answer) as the determining factor for water displacement, thusequivalent to the level of scientific explanations.

Answers on multiple-choice and true-false items of the content area of“explanations of floating and sinking” were grouped together as three sumscores reflecting different levels of conceptual understanding (Misconcep-tions, MC; Explanations of Everyday Life, MC; Scientific Explanations,MC). Here, answers on multiple-choice items of “explanations of specificsituations” were assigned to one of the three scores of conceptual under-standing, depending on which level of conceptual understanding theyexpressed. For example, Answers 2 and 3 on the sample item in Figure 2would be assigned to the score of misconceptions. Answers on the groupof “generalized concepts” were also assigned to one of the three scores ofconceptual understanding, with the items testing material kind as part of thescore of Explanations of Everyday Life. Reliability analyses of scalesshowed satisfactory results, with Cronbach’s alpha (pretest, posttest,follow-up test)/Misconceptions � .59, .62, .75, respectively; Cronbach’salpha (pretest, posttest, follow-up test)/Explanations of Everyday Life �.57, .62, .63, respectively; and Cronbach’s alpha (pretest, posttest,follow-up test)/Scientific Explanations � .70, .78, .81, respectively. Themaximum scores for the three multiple-choice scores were 17 points forScientific Explanations, 11 points for Explanations of Everyday Life, and32 points for Misconceptions.

Answers on free-response items were summed up on four levels ofunderstanding, with separate sum scores for Misconceptions (FR), Expla-nations of Everyday Life (FR), Scientific Explanations 1 (FR), and Scien-tific Explanations 2 (FR). Scientific Explanations 1 includes answers thataddress only one physical quantity, while answers scored as ScientificExplanations 2 integrate two or more physical quantities. Note that for thescores of Scientific Explanations and Explanations of Everyday Life weexpect an increase due to time and instruction, while for the scores ofMisconceptions we expect a decline.

2) Answering pattern on the entire test. Although students’ construc-tion of new scientific explanations and their rejection of or adherence tomisconceptions may be seen as separate processes, it is important to assessthe extent to which students were able to integrate their new explanationsinto a coherent conceptual framework. Therefore, we computed a score thatreflects students’ integrated conceptual understanding of floating and sink-ing, which is based on students’ simultaneous correct rejection of miscon-ceptions and adoption of new scientific explanations. Students’ responsepatterns on all multiple-choice items and true-false items of the contentareas of water displacement and explanations of floating and sinking aswell as the answers on the three free-response items on explanations offloating and sinking were combined to a sum score of Integrated Concep-tual Understanding (ICU). On multiple-choice items of explanations ofspecific situations, it is only if none of the misconceptions and at least oneof the scientific answers within one item have been checked that thestudent scores a point. That is, students scored a point on these items onlyif they showed evidence of adhering to a scientifically acceptable concep-tual framework that does not include the simultaneous inclusion of mis-conceptions. In the score of ICU, the explanations of everyday life are notbeing considered since, although they are correct, they may or may not becombined with scientific explanations. Similarly, students scored a pointon free-response items if they had produced an explanation on Level 3(scientific explanations) without mentioning a misconception (Level 1) atthe same time. Again, Level 2 explanations were not considered in thisscoring.

On the items of water displacement and generalized concepts, studentsneeded to show the correct answering pattern for the entire item context(e.g., the context of material kind) to score one point. Scoring all of the 36items of the test in this stringent way resulted in a maximum of 17 points.

315EFFECTS OF INSTRUCTIONAL SUPPORT

Page 10: Effects of Instructional Support Within Constructivist Learning … · 2009-03-18 · Effects of Instructional Support Within Constructivist Learning Environments for Elementary School

Cronbach’s alpha (pretest, posttest, follow-up test)/ICU (MC) was .48, .82,.81, respectively. Due to the stringent scoring procedure, the variance in thepretest, where students are often confronted with explicit explanations offloating and sinking for the first time, is more restricted than in the posttestand follow-up test.

Transfer Test

In order to test students’ ability to employ the scientific principlesunderlying objects’ floating and sinking in fluids for the explanation ofsituations beyond those covered in instruction, we designed a transfer testwith a total of eight items. This tests relies less on verbal expressions thanthe Test on Floating and Sinking used in the repeated measures design andthus serves as a further indicator of students’ conceptual understanding offloating and sinking.

Test Content

1) Application of the concept of density. A total of five multiple-choice items tested students’ conceptual understanding of density by fo-cusing on the comparison of densities of one object and two liquids or elseof two liquids, thus going beyond the comparison of densities of object andwater assessed in the Test on Floating and Sinking. The application of theconcept of density may be assessed by items requiring little verbal expli-cation for the answers. Rather, students’ conceptual understanding ofdensity may be inferred from their consideration of the correct variablesunderlying a comparison of densities. For example, one item asked stu-dents to determine how much a cube of a given material will weigh if itsinks in oil but floats in water, given the masses of a cube of water and acube of oil, respectively, of the same size as the object. For the correctanswer, students need to know that the number for the mass of the objectneeds to be smaller than the mass of the cube of water and larger than the

mass of the cube of oil (see Figure 4). During the instruction, “a cube ofwater” had been referred to as the amount of water displaced by a cube ofa certain material.

Another item asked students to check which one of four suggestions wouldhelp someone intending to “let an egg float in salty water”: the addition ofmore water, the addition of more salt (correct answer), the addition of more ofthe same salty water, or the use of a smaller egg than the one displayed in theitem. Figure 5 displays an item in which students had to decide which of thedisplayed cubes would sink in water, given the mass and volume of a cube ofwater. In this item, multiple answers could be checked.

2) Application of the concept of buoyancy force. Three multiple-choiceitems focused on students’ explanations of floating and sinking with afocus on buoyancy force. The situations described in the items wentbeyond single objects’ floating and sinking as covered during the curric-ulum by either involving the comparison of two floating objects or thecomparison of two states of an object. Similar to the group of multiple-choice items on explanations of floating and sinking in the Test on Floatingand Sinking, students had to check all explanations they judged as beingcorrect for a specific situation, with possible answers covering misconcep-tions and scientific explanations. Specifically, one item asked students todecide which of two depicted boats of the same mass but different volumewould be able to carry a treasure better and to justify their response.Another item asked students to choose answers which best explained whya piece of clay on a fishing rod feels lighter when immersed in water. Thethird item asked students to choose answers that best explained why fish inwater can rise to the top when they increase their gas bladder (see Figure6). All three items allowed multiple answers to be checked.

Test Administration

The Transfer Test was administered within three days after the firstposttest, that is, within two weeks after the instructional unit. The teacher

Figure 4. Sample item from the Transfer Test on the application of density.

316 HARDY, JONEN, MOLLER, AND STERN

Page 11: Effects of Instructional Support Within Constructivist Learning … · 2009-03-18 · Effects of Instructional Support Within Constructivist Learning Environments for Elementary School

proceeded the same way as she had done for the Test on Floating andSinking, explaining all of the situations and item formats and demonstrat-ing all of the experimental situations described in the test booklet using realobjects and material (such as oil floating on water). Then students pro-ceeded to work on the test booklets independently in a prescribed sequenceof items, taking their individually needed time.

Scoring

We computed separate sum scores for misconceptions and scientificexplanations. Since items on the Transfer Test focused on students’ appli-cation of the concepts of density and buoyancy force in a wider context, noanswers involving explanations of everyday life were provided. The sumscore for Scientific Explanations (TR) was computed by summing upcorrect answers of all eight items, with a maximum of 19 points and � �.53. The correlations of this score with the sum score of Scientific Expla-nations (MC) of the Test on Floating and Sinking are .46 (posttest) and .32(follow-up test). Similarly, the sum score for Misconceptions (TR) consistsof the total of adopted misconceptions provided in the eight items of theTransfer Test, with a maximum of 18 points and � � .54. The correlationsof this score with the sum score of Misconceptions (MC) of the Test onFloating and Sinking are .58 (posttest) and .33 (follow-up test). Since halfof the items of the Transfer Test were true multiple-choice items with onlyone correct answer, we did not compute a score of ICU. The mediumcorrelations between scores of both tests for each conceptual level (Mis-conceptions and Scientific Explanations) suggest that the answers providedin multiple-choice items of the Test on Floating and Sinking may beregarded as a valid indicator of students’ conceptual understanding offloating and sinking.

Implementation Checks

Because both instructional units were administered by the same teacher,it had to be ensured that the teacher actually implemented the characteristic

dimensions of the instructional variation. We employed two implementa-tion checks to ensure the experimental variation of HIS versus LIS in eachof the six experimental classes.

(1) Each of the 48 videotaped 90-min lessons was rated by a total of 12independent raters as either a lesson of HIS or LIS. As criteria, the raters hadreceived a list of characteristic elements of instruction intended for the twocurricula, which were based on the two manipulated domains of the sequenc-ing of content and the teacher’s cognitive structuring of classroom discourse.

(2) Thirty percent of tapes were randomly chosen to be categorizedaccording to the types of verbal utterances by students and teachers duringclass discussions, thus excluding independent work at the experimentallearning sites. The drawing of videos ensured that all of the lessons in theinstructional sequence and all of the classes were represented equally.Verbal utterances were categorized in 10-s sequences, with the possibilityof multiple codes per unit. Teacher statements were coded on whether theyactively structured conversation on a content level, for example, by addressingmisconceptions or requesting student explanations; whether they passivelykept the conversation going, for example, by repeating student statements; orwhether they concerned managerial issues. Note that none of the statementsintended as structural comments concerned explicit explanations on a contentlevel. Rather, they intended to focus students’ attention on relevant aspects ofinvestigated phenomena and to model reasoning about scientifically relevantrelationships. Student statements were rated as to their active contribution ofcontent, their contribution of content as a reaction to a teacher statement, andtheir contribution of content as a reaction to another student’s statements. Thecoding scheme was exhaustive regarding teacher and student statements. TheAppendix summarizes the specifications of the coding scheme for the rating ofteacher and student discourse.

RESULTS

Implementation Checks

For our first implementation of the overall categorization oflessons as either part of the curriculum of HIS or LIS, the interrater

Figure 5. Sample item from the Transfer Test on the application of density.

317EFFECTS OF INSTRUCTIONAL SUPPORT

Page 12: Effects of Instructional Support Within Constructivist Learning … · 2009-03-18 · Effects of Instructional Support Within Constructivist Learning Environments for Elementary School

reliability was .94 with a master coder. For our second implemen-tation check of ratings of teachers’ and students’ verbal statementsduring class discussions, we performed a multivariate analysis ofvariance (MANOVA) for the entirety of coded statements at 10-sintervals and chi-square tests for single codes. There was a signif-icant effect for group differences at the multivariate level, Wilks’s� � .83, F(7, 2821) � 82.24, p � .001, �2 � .17, with significantdifferences between the LIS and HIS environments for all of thecoded statements using chi-square tests. Table 3 summarizes de-scriptive statistics and results of chi-square tests for single codes.It is important to note that teachers in the HIS condition gavecognitively structuring statements twice as frequently as did teach-ers in the LIS condition (relative to the total number of coded 10-sintervals), while they provided managerial comments about half asoften as did teachers in the LIS classrooms. As expected, studentsin the LIS condition showed higher relative frequencies for reac-tions to other students’ contributions and for independent content

contributions than students in the HIS condition, while students inthe HIS condition more often reacted to teacher statements.

Overall, our two implementation checks underline that in the sixexperimental classrooms the intended variation of instructionalsupport with regard to the sequencing of content and the structur-ing of social interaction was implemented according to the de-signed curricula.

Students’ Achievement Gains on the Test on Floating andSinking

In a first step, differences between the two experimental groupsand the baseline group were examined with regard to students’overall answering pattern on the Test on Floating and Sinking onthe pretest, posttest, and follow-up test. In order to explain andfurther analyze overall differences in answering patterns, we thenproceeded with analyses of separate conceptual levels of under-

Figure 6. Sample item from the Transfer Test on the application of buoyancy force.

318 HARDY, JONEN, MOLLER, AND STERN

Page 13: Effects of Instructional Support Within Constructivist Learning … · 2009-03-18 · Effects of Instructional Support Within Constructivist Learning Environments for Elementary School

standing across the three points of measurement. In the followingsections, we first report on tests to identify possible pretest differ-ences between groups. Then the results of overall tests of meancomparison are reported, followed by tests on differences betweenpoints of measurement (pretest to posttest; pretest to follow-uptest). We first address differences between instructional groups andthen between instructional groups and the baseline group. If ap-propriate, results of t test pertaining to gains within groups are alsoreported. The chosen � level is .05.

Integrated Conceptual Understanding of Floating andSinking

A univariate analysis of variance ensured that there were nopretest differences between the groups of HIS, LIS, and the base-line group (BS) on the score of ICU, F(2, 158) � 0.24, ns. To testthe effects of instructional support on students’ conceptual under-standing of floating and sinking, we performed a 3 (time) � 3(group) ANOVA with the score of ICU. Figure 7 illustrates meandifferences across points of measurement for each group. Therewere significant effects of time, F(2, 158) � 104.25, p � .001,�2 � .40; Time � Group, F(4, 316) � 10.09, p � .001, �2 � .11;and group, F(2, 158) � 12.46, p � .001, �2 � .14.

Follow-up ANOVAS with scores of differences between pointsof measurement (pretest to posttest; pretest to follow-up test)revealed that the two instructional groups did not significantly

differ from each other in their gains from pretest to posttest;however, they significantly differed from the baseline group (ef-fect of instruction: F(2, 158) � 15.27, �2 � .16, p � .001; plannedcontrast HIS–BS: p � .001, d � 1.24; planned contrast LIS–BS:p � .001, d � 1.11). In a comparison of gains from pretest tofollow-up test, there was a significant effect for group, F(2, 158) �10.49, p � .001; �2 � .12, with a significant difference betweengains of the two instructional groups ( p � .05, d � .38), as wellas between the two instructional groups and the baseline group(planned contrast HIS–BS: p � .001, d � .98; planned contrastLIS–BS: p � .05, d � .61).

T tests of mean values for each group revealed that while thegroup of HIS did not change significantly from posttest tofollow-up test, there was a significant decline of mean scores in thegroup of LIS, t(58) � 2.29, p � .05, d � .20. T tests furthershowed that the baseline group also gained significantly frompretest to posttest, t(35) � �2.65, p � .05, d � .54, and in the longrun in the follow-up measure, t(35) � �3.30, p � .01, d � .71.However, the gains from pretest to follow-up test for the group ofHIS and the group of LIS were larger (HIS: t(66) � �10.61, p �.001, d � 1.75; LIS: t(59) � �7.94, p � .001, d � 1.32). In sum,analyses of the score of ICU showed a significant advantage of thegroup of HIS as compared to the group of LIS in the follow-up test,while both experimental groups showed significantly higher meansthan the baseline groups both at the posttest and the follow-up test.

Figure 7. Means of score of Integrated Conceptual Understanding (ICU) by time and group.

Table 3Results of Chi-Square Tests of Teacher and Student Statements at 10-s Intervals

Variable Statement type

Relative frequencies

�2 p dHIS

(N � 1,567)LIS

(N � 1,262)

Teacher Structuring .63 .28 337.83 .000 .76Passive .12 .08 10.61 .001 .14Managerial .15 .29 85.43 .000 .35

Student Active .06 .15 55.0 .000 .37Reaction to student .06 .20 120.02 .000 .23Reaction to teacher .48 .33 71.11 .000 .31Other .04 .05 9.36 .002 .05

Note. HIS � high instructional support; LIS � low instructional support.

319EFFECTS OF INSTRUCTIONAL SUPPORT

Page 14: Effects of Instructional Support Within Constructivist Learning … · 2009-03-18 · Effects of Instructional Support Within Constructivist Learning Environments for Elementary School

Analyses of Separate Conceptual Levels

For further investigation of the composition of answering pat-terns combined within the score of ICU, we performed analyseswith eight separate scores of conceptual understanding separatelyfor multiple-choice and free-response items.

Univariate ANOVA for pretest differences between groupsshowed no significant differences for the scores of Water Dis-placement, Misconceptions (FR, MC), Explanations of EverydayLife (FR, MC), Scientific Explanations 1 (FR), and ScientificExplanations 2 (FR). There was a significant difference in meanpretest scores of Scientific Explanations (MC), F(2, 158) � 3.36,p � .05, �2 � .04. Scheffe post hoc comparisons showed adifference between the groups of HIS and LIS ( p � .051), with agreater mean of the group of HIS, M (HIS) � 7.20, M (LIS) �5.73.

A 3 (time) � 3 (group) MANOVA with the dependent measuresof Water Displacement, Misconceptions (MC, FR), Explanationsof Everyday Life (MC, FR), Scientific Explanations (MC), Scien-tific Explanations 1 (FR), and Scientific Explanations 2 (FR)resulted in significant effects of time, Wilks’s � � .21, F(16,143) � 31.22, p � .001, �2 � .78; time � group, Wilks’s � � .44,F(32, 286) � 4.58, p � .001, �2 � .34; and group, Wilks’s � �.57, F(16, 302) � 5,90, p � .001, �2 � .24. Tables 4 and 5 listmeans and standard deviations for the multiple-choice items andfree-response items, respectively, for the three experimentalgroups at three points of measurement.

Univariate follow-up analyses with each of the dependent mea-sures were performed. The respective statistics are listed in Tables6 and 7. There were significant gains across points of measurementfor all dependent variables, and there were significant interactionsof Time � Group for all dependent variables except Explanationsof Everyday Life (MC) and Water Displacement (MC). Table 8lists results of ANOVAs for differences between points of mea-surement (pretest–posttest and pretest–follow-up test), which wereperformed to determine the difference in gains between the threegroups relevant to our hypotheses.

Differences From Pretest to Posttest

As may be seen from Table 8, the two instructional groups showsignificant differences only in their gains on Scientific Explana-tions 1 (FR). The group of HIS outperformed the baseline group inall measures except for the scores of Explanations of EverydayLife (MC, FR). The group of LIS showed a similar pattern;however, they also outperformed the baseline group with regard tothe production of Explanations of Everyday Life (FR), while theydid not differ from them in the production of Scientific Explana-tions 1 (FR). The differences in the reduction of misconceptionsand the adoption of scientific explanations between the two ex-perimental groups and the baseline group can be considered verylarge, with Cohen’s d ranging from .89 to 1.71.

T tests for single groups showed that the baseline group did notchange significantly with regard to mean scores of Water Displace-ment, Misconceptions (MC), Explanations of Everyday Life (FR),and Scientific Explanations (FR); however, there was a significantdecrease from pretest to posttest on free-response answers rated asMisconceptions (FR), t(35) � 2.84, p � .01, d � .36, as well as anincrease on Explanations of Everyday Life (MC), t(35) � �3.4, p � T

able

4M

eans

(and

Stan

dard

Dev

iati

ons)

ofSe

para

teL

evel

sof

Con

cept

ual

Und

erst

andi

ng(M

ulti

ple-

Cho

ice

Item

s)by

Tim

ean

dG

roup

Con

tent

Con

cept

ual

leve

l

Pret

est

Post

test

Follo

w-U

pte

st

HIS

(N�

66)

LIS

(N�

59)

BS

(N�

36)

HIS

LIS

BS

HIS

LIS

BS

Wat

erD

ispl

acem

ent

Scie

ntif

icE

xpla

natio

ns(M

C)

3.46

(2.3

5)3.

63(2

.10)

3.47

(2.5

6)5.

24(3

.15)

5.41

(3.2

0)4.

03(2

.69)

5.65

(2.9

1)4.

83(3

.20)

4.58

(3.1

7)E

xpla

natio

nsof

Floa

ting

and

Sink

ing

Mis

conc

eptio

ns(M

C)

14.2

1(3

.70)

12.9

3(3

.94)

13.0

8(3

.94)

8.48

(5.0

5)8.

85(5

.04)

13.2

5(4

.28)

8.62

(4.4

3)9.

14(4

.36)

12.0

8(4

.20)

Exp

lana

tions

ofE

very

day

Lif

e(M

C)

7.18

(2.3

0)6.

51(1

.87)

6.72

(2.3

9)9.

21(1

.70)

8.54

(2.0

6)8.

03(1

.80)

8.07

(2.2

5)7.

69(1

.98)

8.25

(1.8

6)Sc

ient

ific

Exp

lana

tions

(MC

)7.

20(3

.21)

5.73

(3.5

0)7.

11(3

.23)

11.4

7(3

.88)

10.4

1(3

.65)

8.36

(3.3

1)10

.29

(4.4

6)9.

37(3

.91)

8.58

(3.1

0)

Not

e.H

IS�

high

inst

ruct

iona

lsu

ppor

t;L

IS�

low

inst

ruct

iona

lsu

ppor

t;B

S�

base

line

grou

p;M

C�

mul

tiple

choi

ce.

320 HARDY, JONEN, MOLLER, AND STERN

Page 15: Effects of Instructional Support Within Constructivist Learning … · 2009-03-18 · Effects of Instructional Support Within Constructivist Learning Environments for Elementary School

.01, d � .63, and on Scientific Explanations (MC), t(35) � �2.43,p � .05, d � .38. Thus, the experience of test taking apparently led toconceptual advances even in the baseline group.

Differences From Pretest to Follow-Up Test

There were significant differences with regard to the decline ofMisconceptions (MC, FR) and the gain in Scientific Explanations2 (FR) between the two instructional groups. At the follow-up test,the group of HIS showed significantly fewer misconceptions bothin multiple-choice and free-response items than did the group ofLIS, and students produced significantly more answers rated asScientific Explanations 2. Generally, differences between instruc-tional groups with regard to misconceptions seem to have in-creased over time, while differences with regard to the productionof scientific explanations seem to have shifted from a focus ononly one quantity (Scientific Explanations 1) at the posttest to afocus on two or more quantities (Scientific Explanations 2) at thefollow-up test.

T tests within groups revealed that the group of LIS showed asignificant increase from posttest to follow-up measure on thescores of Misconceptions (FR), t(58) � �3.26, p � .01, d � .47;thus, students regained many of the misconceptions they hadalready resolved a year earlier at the posttest. The group of HIS, incontrast, did not change significantly from posttest to follow-uptest with regard to the production or adoption of misconceptions.

With regard to answers rated as Scientific Explanations 2, thegroup of LIS showed significant gains from pretest to follow-uptest, t(58) � �6.15, p � .001, d � 1.36, as did the group of HIS,t(65) � �7.20, p � .001, d � 1.41. The baseline group did notproduce any answers rated as Scientific Explanations 2 at anypoint of measurement. The groups of LIS and HIS showed asignificant increase in answers rated as Scientific Explanations 1(FR) (LIS: t(58) � �6.67, p � .01, d � .79; HIS: t(65) � �3.78,p � .001, d � .69), while the baseline group did not significantlyimprove from pretest to follow-up test.

In comparisons of pretest and follow-up test scores, the twoinstructional groups and the baseline group show similar gains onscores of Explanations of Everyday Life (FR, MC), with Cohen’sd ranging from .14 to .31. On the score of Scientific Explanations1 (FR), the groups of HIS and LIS still outperform the baselinegroup. On the score of Water Displacement (MC), only the groupof HIS shows greater gains than the baseline group. T tests forsingle groups showed that from pretest to follow-up test bothinstructional groups and the baseline group gained significantly onExplanations of Everyday Life (MC) (HIS: t(65) � �2.50, p �.05, d � .39; LIS: t(65) � 4.47, p � .001, d � .62; BS: t(35) ��3.65, p � .001, d � .73). On free-response items of Explanationsof Everyday Life (FR), however, gains are significant from pretestto posttest for only the two instructional groups (HIS: t(65) ��3.01, p � .001, d � .53; LIS: t(58) � �4.76, p � .001, d � .86),

Table 5Means (and Standard Deviations) of Separate Levels of Conceptual Understanding (Free-Response Items) by Time and Group

Conceptual level

Pretest Posttest Follow-up test

HIS(N � 66)

LIS(N � 66)

BS(N � 66) HIS LIS BS HIS LIS BS

Misconceptions (FR) 2.62 (1.29) 2.32 (1.06) 2.58 (1.23) .76 (1.04) .86 (1.01) 2.14 (1.20) .79 (.89) 1.37 (1.16) 1.92 (.97)Explanations of

Everyday Life (FR) 1.04 (1.33) 1.32 (1.06) 1.11 (1.33) 1.71 (1.19) 2.46 (1.61) 1.19 (1.08) 1.35 (1.13) 1.49 (1.33) 1.72 (1.22)Scientific Explanations

1 (FR) .26 (.55) .27 (.54) .08 (.26) .85 (1.03) .44 (.58) .28 (.60) .82 (1.04) .86 (.94) .31 (.61)Scientific Explanations

2 (FR) .05 (.26) .02 (.12) .00 (.00) 1.21 (1.44) .92 (1.06) .00 (.00) 1.08 (1.18) .66 (.81) .00 (.00)

Note. HIS � high instructional support; LIS � low instructinal support; BS � baseline group; FR � free response.

Table 6Analyses of Variance for Levels of Conceptual Understanding (Multiple-Choice Items)

Source

Water Displacement MisconceptionsExplanations of Everyday

Life Scientific Explanations

df F �2 p F �2 p F �2 p F �2 p

Between subjects

Group 2 1.31 .016 .27 6.93 .081 .001 2.61 .032 .077 4.35 .052 .014Error 158

Within subjects

Time 2 32.28 .128 .000 51.14 .245 .000 42.47 .212 .000 65.53 .293 .000Time � Group 4 2.31 .028 .058 10.35 .116 .000 1.94 .024 .103 5.06 .060 .001Error 316

321EFFECTS OF INSTRUCTIONAL SUPPORT

Page 16: Effects of Instructional Support Within Constructivist Learning … · 2009-03-18 · Effects of Instructional Support Within Constructivist Learning Environments for Elementary School

while the baseline group also showed gains in a comparison ofpretest and follow-up test, t(35) � 3.24, p � .001, d � .48. Withregard to Scientific Explanations 1 (FR), t tests for single groupsshowed that while both instructional groups gained from pretest tofollow-up test (HIS: t(65) � 10.21, p � .001, d � .69; LIS: t(58) �7.44, p � .001, d � .80), the baseline group did not showsignificant changes on this variable.

To sum up analyses of scores of separate conceptual levels,there was a clear advantage of a high degree of instructionalsupport with regard to students’ more frequent production ofscientific explanations with respect to one quantity in the posttest.The long-term effects of instructional support were evident withregard to both misconceptions and scientific explanations. In thefollow-up test, students of the group of HIS significantly differedfrom students of the group of LIS on the less frequent selection andproduction of misconceptions and the more frequent production ofscientific explanations with respect to two or more scientificconcepts. While from posttest to follow-up test the group of LISsignificantly increased their production of misconceptions, themeans of the group of HIS did not change. Additionally, there is along-term impact of instruction on the score of Water Displace-ment, where the group of HIS outperformed both the group of LISand the baseline group. It is interesting that, over the long run, thetwo instructional groups did not differ from the baseline group on

both the production and selection of Explanations of EverydayLife. The differences between the groups witnessed in theoverall answering pattern on the score of ICU are apparentlyproduced by differences with regard to Misconceptions (MC,FR) and Scientific Explanations 2 (FR) in the follow-up test,while in the posttest the superiority on Scientific Explanations1 (FR) in the group of HIS did not contribute enough varianceto affect the overall score of ICU.

Group Differences on the Transfer Test

For the Transfer Test, separate sum scores of Misconceptionsand Scientific Explanations were computed. Means and standarddeviations for the three groups were M (HIS) � 7.03 (SD � 3.22),M (LIS) � 7.35 (SD � 3.26), and M (BS) � 10.42 (SD � 2.72)for the score of Misconceptions and M (HIS) � 13.32 (SD � 2.96),M (LIS) � 12.00 (SD � 2.70), and M � 8.78 (SD � 2.35) for thescore of Scientific Explanations. A MANOVA with the factor ofgroup (LIS, HIS, BS) and the scores of Misconceptions andScientific Explanations showed a significant effect of group,Wilks’s � � .68, F(4, 312) � 16.31, p � .001, �2 � .17.Univariate ANOVAs revealed a significant group effect for bothdependent variables, with F(2, 157) � 32.13, p � .001, �2 � .29for Scientific Explanations and F(2, 157) � 15.08, p � .001, �2 �

Table 7Analyses of Variance for Levels of Conceptual Understanding (Free-Response Items)

Source

MisconceptionsExplanations of Everyday

Life Scientific Explanations 1 Scientific Explanations 2

df F �2 p F �2 p F �2 p F �2 p

Between subjects

Group 2 12.97 .141 .000 3.72 .045 .026 9.82 .111 .000 17.77 .184 .000Error 158

Within subjects

Time 2 86.32 .35 .000 12.23 .072 .000 14.58 .084 .000 42.06 .210 .000Time � Group 4 9.95 .112 .000 5.40 .064 .000 2.53 .031 .041 10.24 .115 .000Error 316

Table 8P Values and Cohen’s d for Planned Contrasts With Mean Gains Between Points of Measurement

Conceptual level

Pretest–posttest Pretest–follow-up test

LIS HIS HIS BS LIS BS LIS HIS HIS BS LIS BS

p d p d p d p d p d p d

Water Displacement (MC) .98 .01 .01 .45 .01 .46 .08 .32 .09 .34 .89 .03Misconceptions (MC) .07 .29 .01 1.15 .01 .89 .03 .39 .01 .97 .01 .58Misconceptions (FR) .07 .21 .01 1.28 .01 .89 .01 .59 .01 .90 .35 .22Explanations of Everyday Life (MC) .99 .01 .11 .33 .12 .33 .55 .11 .24 .23 .54 .14Explanations of Everyday Life (FR) .12 .26 .09 .45 .01 .80 .64 .07 .36 .22 .19 .31Scientific Explanations (MC) .48 .12 .01 .93 .01 1.07 .48 .13 .07 .38 .02 .51Scientific Explanations 1 (FR) .02 .42 .05 .43 .90 .03 .87 .02 .14 .36 .11 .40Scientific Explanations 2 (FR) .18 .23 .01 1.65 .01 1.71 .02 .40 .01 1.76 .01 1.60

Note. HIS � high instructional support; LIS � low instructional support; BS � baseline group; MC � multiple choice; FR � free response.

322 HARDY, JONEN, MOLLER, AND STERN

Page 17: Effects of Instructional Support Within Constructivist Learning … · 2009-03-18 · Effects of Instructional Support Within Constructivist Learning Environments for Elementary School

.16 for Misconceptions. Planned contrasts confirmed a significantdifference between the instructional groups of HIS and LIS and thebaseline group (HIS–BS: p � .001, d � 1.71; LIS–BS: p � .001,d � 1.28) and significantly higher scores for the HIS groupcompared to the LIS group ( p � .01, d � .47) for the score ofScientific Explanations. For the score of Misconceptions, the twoinstructional groups differed significantly from the baseline groupbut not from each other (HIS–BS: p � .001, d � 1.14; LIS–BS:p � .001, d � 1.03).

DISCUSSION

The major concern of this study was to contribute to the ques-tion of how best to support elementary school students’ conceptualunderstanding of complex science topics within classroom envi-ronments based on constructivist ideas of learning. Our learningenvironments allowed for discovery learning by hands-on experi-ence with authentic material and encouraged sensemaking throughpeer interaction and classroom discussion, yet differed in thedegree of instructional support. Under the condition of LIS, stu-dents could themselves decide when and how long to make use ofexperimental material within a given time frame, and the class-room discussions were student centered. Under the condition ofHIS, students’ experimental activities were sequenced into activ-ities addressing basic concepts such as material kind to addressingmore complex concepts such as the relationship between buoyancyforce and water displacement. During classroom discussions, theteacher repeatedly requested clarifications and justifications ofstudent ideas, addressed misconceptions by empirical evidence, orfostered processes of generalization (Reiser, 2004).

In accordance with conjectures put forward by Mayer (2004),our results showed that elementary school students’ conceptualunderstanding of floating and sinking can be optimized by instruc-tional support provided through the sequencing of instructionalcontent and the frequency of cognitively structuring statements bythe teacher. In the long run, students in the group of HIS developeda more coherent understanding of why some objects float in waterwhile others sink than did students in the group of LIS. Withregard to the score of ICU, students in the group of HIS could keeptheir level of conceptual understanding from posttest to follow-uptest, while students in the group of LIS showed a significantdecline. Among students in the group of LIS, misconceptionsreemerged to a significant extent between posttest and follow-uptest, while mean values in the group of HIS did not change. Similarpatterns were also found for the scores of Water Displacement andScientific Explanations 2, where two of three physical quantities—density, water displacement, or buoyancy force—must be takeninto account simultaneously. A comparison of gains in producingscientific explanations from pretest to follow-up test revealed nodifferences between both instructional groups for Scientific Expla-nations 1, but larger gains in the group of HIS for ScientificExplanations 2. The superiority of students in the group of HISwith regard to the integration of physical concepts in scientificexplanations is also indicated by their higher scores on the Trans-fer Test where the concepts of density and buoyancy force neededto be applied in new contexts.

Is it conceivable that our test scores really reflect a morecoherent conceptual understanding in students of the HIS group?Several arguments may be put forward in support of this claim.

First, the correlations between our separate scores of conceptualunderstanding on the Test of Floating and Sinking and the TransferTest, assessing the quality of students’ explanations for contextsnot treated during the curriculum, suggest that we assessed asimilar construct in both tests. Second, the variety of students’explanations on free-response items for explanations scored asScientific Explanations 1 and 2 shows that students did not mem-orize fragments of explanations provided during instruction. Hadstudents actually memorized rules, the superiority of the group ofHIS in the follow-up test with regard to Scientific Explanations 2would have been highly unlikely. In addition, our implementationcheck shows that although the teacher focused students’ attentionon relevant aspects of investigated phenomena under the conditionof HIS, she never explicated scientific rules herself. Thus, thelearners of the group of HIS did not get better opportunities forrote-leaning teacher statements than students in the group of LIS.Third, conceptual understanding was assessed not only in terms ofthe production of scientific explanations but also in terms of adecline in the production of misconceptions. The group of HIS wasclearly superior in this aspect.

The relatively small differences between both instructionalgroups in the posttest and the clear superiority of the group of HISone year later suggest that different kinds of conceptual changeoccurred during the period in which the curriculum was appliedand during the following year when no instruction on floating andsinking was encountered. By running experiments and by inter-acting with their peers, learners from both groups restructured theirknowledge about water displacement, extended their repertoire ofexplanations, and learned that some of their previously constructedexplanations were inappropriate. However, while students in thegroup of HIS integrated the knowledge acquired during the cur-riculum into a coherent conceptual system where inappropriateexplanations would be replaced by scientific ones, learners in thegroup of LIS may not have contrasted their existing misconcep-tions with new explanations to a sufficient degree (see Davis,2003; Linn, 1995). Thus, a reemergence of misconceptions wasobserved in the follow-up test, although students still rememberedthe scientific explanations adopted during the curriculum, as thehigh scores on the multiple-choice items and the production ofScientific Explanations 1 indicate. In other words, under the con-dition of LIS learners may not have realized the incompatibility ofdifferent explanations to the same degree as did learners under thecondition of HIS. When working on the follow-up test, the expla-nations developed during the curriculum competed with the earliermisconceptions, thus leading to an inconsistent response pattern.This pattern of reoccurring misconceptions is commonly found inresearch on conceptual change, indicating that previously adoptedexplanations were not well integrated into existing conceptualframeworks (e.g., Vosniadou et al., 2001).

In the current design, the sequencing of content and the struc-turing of classroom discussions are confounded; therefore, thesingle contribution of each dimension remains as yet unknown.Whether there are additive effects of both dimensions, or whethereither dimension is sufficient for inducing conceptual change, mustbe clarified by further studies with experimental variations ofstructuring classroom practice. It is also plausible that the interac-tion between sequenced content and cognitively structuringteacher statements produced the effectiveness of the HIS environ-ment for conceptual change. That is, the sequencing of content

323EFFECTS OF INSTRUCTIONAL SUPPORT

Page 18: Effects of Instructional Support Within Constructivist Learning … · 2009-03-18 · Effects of Instructional Support Within Constructivist Learning Environments for Elementary School

probably led to classroom discussions that were focused on spe-cific aspects of investigations experienced by all students of aclass. Therefore, the teacher may have been able to provide cog-nitively structuring statements in response to individual studentconjectures and explanations which in effect also targeted the levelof conceptual understanding of most students in the class. Incontrast, the teacher’s structuring statements in the group of LIS,occurring with significantly lower frequency, needed to addressindividual students’ ideas to a greater degree, due to the variety ofexperiments and content on which students had worked duringprior independent work. It is likely that these teacher statementswere not as effective in furthering the conceptual understanding ofstudents who had worked on different instructional material priorto the discussion.

In an extension of Hogan and Pressley (1997), we suggest thatthe provision of scaffolding need not be an intervention based onindividual tutoring but may be implemented within an entireclassroom; the sequencing of content, contributing to the similarityof students’ conceptual background, will likely add to the effec-tiveness of teacher-provided scaffolding. In our study, elementaryschool students with little prior domain-specific knowledge prof-ited from participation in a learning environment with instructionalsupport. Further research needs to address the extent to whichlearners of different age groups and cognitive preconditions suchas greater metacognitive control will profit from scaffolded learn-ing environments to a similar degree.

CONCLUSIONS

Why should children as young as nine years of age learn aboutthe concepts of density and buoyancy force? Data from our base-line group showed that the adoption of answers based on expla-nations of everyday life is a conceptual advancement during ele-mentary school that does not need to be supported by instruction toa great extent. In contrast, conceptual advances with regard to thereduction of misconceptions and the adoption and construction ofscientific explanations do need to be supported by instruction.Directing young children’s attention toward basic physics expla-nations may lay the foundation for their later understanding at amore formal level. In addition, participation in learning environ-ments of challenging science content will likely support the de-velopment of students’ scientific reasoning skills such as theirexperimental skills, their use of empirical data to support theoriz-ing, and their ability to reason about patterns of covariation (e.g.,Bullock & Ziegler, 1999; Toth, Klahr, & Chen, 2000).

In our curriculum, children were directed to discover that ifobjects are immersed into water, water pushes against the object.This understanding may support the understanding that also forcesalways come in pairs—equal and opposite forces of action andreaction. It is well documented that even after having participatedin physics instruction in secondary school students rarely under-stand the concept of force as a characteristic of a system, but rathertreat it as a characteristic of single objects (McCloskey, 1983). Itneeds to be left to future longitudinal studies to show that an earlyunderstanding of science concepts in elementary school will helpstudents to profit better from learning environments provided insecondary education.

References

Bransford, J., Brown, A., & Cocking, R. (Eds.). (1999). How people learn.Washington, DC: National Academy Press.

Bullock, M., & Ziegler, A. (1999). Scientific reasoning: Developmentaland individual differences. In F. E. Weinert & W. Schneider (Eds.),Individual development from 3 to 12 (pp. 38–54). Cambridge, England:Cambridge University Press.

Caravita, S. (2001). A re-framed conceptual change theory? Learning &Instruction, 11, 421–429.

Carey, S. (2000). Science education as conceptual change. Journal ofApplied Developmental Psychology, 21, 13–19.

Chi, M. T. H. (2005). Commonsense conceptions of emergent processes:Why some misconceptions are robust. Journal of the Learning Sciences,14, 161–199.

Chinn, C. A., & Brewer, W. F. (1993). The role of anomalous data inknowledge acquisition: A theoretical framework and implications forscience instruction. Review of Educational Research, 63, 1–49.

Christianson, R. G., & Fisher, K. M. (1999). Comparison of studentlearning about diffusion and osmosis in constructivist and traditionalclassrooms. International Journal of Science Education, 21, 687–698.

Cobb, P. (1994). Where is the mind? Constructivist and socioculturalperspectives on mathematical development. Educational Researcher,23, 13–20.

Davis, E. A. (2003). Prompting middle school science students for pro-ductive reflection: Generic and directed prompts. Journal of the Learn-ing Sciences, 12, 91–142.

Davis, E. A., & Linn, M. C. (2000). Scaffolding students’ knowledgeintegration: Prompts for reflection in KIE. International Journal ofScience Education, 22, 819–837.

Davis, E., & Miyake, N. (2004). Explorations of scaffolding in complexclassroom systems. The Journal of the Learning Sciences, 13, 265–272.

diSessa, A. A. (1988). Knowledge in pieces. In G. Forman & P. B. Pufall(Eds.), Constructivism in the computer age: The Jean Piaget symposiumseries (pp. 49–70). Hillsdale, NJ: Erlbaum.

Greeno, J., & the Middle School Mathematics Through ApplicationsProject Group. (1998). The situativity of knowing, learning, and re-search. American Psychologist, 53, 5–26.

Halford, G. S., Brown, C. A., & Thompson, R. M. (1986). Children’sconcepts of volume and flotation. Developmental Psychology, 22, 218–222.

Hogan, K., & Pressley, M. (1997). Scaffolding scientific competencieswithin classroom communities of inquiry. In K. Hogan & M. Pressley(Eds.), Scaffolding student learning: Instructional approaches and is-sues (pp. 74–107). Louiseville, Quebec, Canada: Brookline Books.

Kohn, A. S. (1993). Preschoolers’ reasoning about density: Will it float?Child Development, 64, 1637–1650.

Lehrer, R., Schauble, L., Strom, D., & Pliggie, M. (2001). Similarity ofform and substance. Modeling material kind. In M. Carver & D. Klahr(Eds.), Cognition and instruction: Twenty-five years of progress (pp.39–74). Mahwah, NJ: Erlbaum.

Linn, M. C. (1995). Designing computer learning environments for engi-neering and computer science: The scaffolded knowledge integrationframework. Journal of Science Education and Technology, 4, 103–126.

Mayer, R. (2004). Should there be a three-strikes rule against pure discov-ery learning? The case for guided methods of instruction. AmericanPsychologist, 59, 14–19.

McCloskey, M. (1983). Naive theories of motion. In D. Gentner & A. L.Stevens (Eds.), Mental models (pp. 299–324). Hillsdale, NJ: Erlbaum.

Moller, K. (1999). Konstruktivistisch orientierte Lehr-Lernprozessfor-schung im naturwissenschaftlich-technischen Bereich des Sachunter-richts [Research of constructivist learning and teaching processes inscience and technical domains of science instruction]. In W. Kohnlein(Ed.), Vielperspektivisches Denken im Sachunterricht [Thinking in mul-

324 HARDY, JONEN, MOLLER, AND STERN

Page 19: Effects of Instructional Support Within Constructivist Learning … · 2009-03-18 · Effects of Instructional Support Within Constructivist Learning Environments for Elementary School

tiple perspectives in science education] (pp. 125–191). Bad Heilbrunn,Germany: Klinkhardt.

Pea, R. (2004). The social and technological dimensions of scaffolding andrelated theoretical concepts for learning, education, and human activity.The Journal of the Learning Sciences, 13, 423–451.

Posner, G., Strike, K., Hewson, P., & Gertzog, W. (1982). Accomodationof scientific conception: Toward a theory of conceptual change. ScienceEducation, 66, 211–227.

Reiser, B. (2004). Scaffolding complex learning: The mechanisms ofstructuring and problematizing student work. The Journal of the Learn-ing Sciences, 13, 273–304.

Smith, C., Carey, S., & Wiser, M. (1985). On differentiation: A case studyof the development of the concepts of size, weight, and density. Cog-nition, 21, 177–237.

Smith, C., Maclin, D., Grosslight, L., & Davis, H. (1997). Teaching forunderstanding. A study of students’ preinstruction theories of matter anda comparison of the effectiveness of two approaches to teaching aboutmatter and density. Cognition and Instruction, 15, 317–393.

Smith, C., Snir, J., & Grosslight, L. (1992). Using conceptual models tofacilitate conceptual change: The case of weight–density differentiation.Cognition and Instruction, 9, 221–283.

Smith, J., diSessa, A., & Roschelle, J. (1993). Misconceptions reconceived:A constructivist analysis of knowledge in transition. The Journal of theLearning Sciences, 3, 115–163.

Staub, F., & Stern, E. (2002). The nature of teachers’ pedagogical contentbeliefs matters for students’ achievement gains: Quasi-experimentalevidence from elementary mathematics. Journal of Educational Psy-chology, 94, 344–355.

Toth, E. E., Klahr, D., & Chen, Z. (2000). Bridging research and practice:A cognitively based classroom intervention for teaching experimentationskills to elementary school children. Cognition & Instruction, 18, 423–459.

Tynjala, P. (1999). Towards expert knowledge? A comparison between acontructivist and a traditional learning environment in university. Inter-national Journal of Educational Research, 31, 357–442.

Tyson, L. M., Venville, G. J., Harrison, A. G., & Treagust, D. F. (1997).A multidimensional framework for interpreting conceptual changeevents in the classroom. Science Education, 81, 387–404.

Vosniadou, S. (1994). Capturing and modeling the process of conceptualchange. Learning and Instruction, 4, 45–69.

Vosniadou, S., & Brewer, W. F. (1992). Mental models of the earth: Astudy of conceptual change in childhood. Cognitive Psychology, 24,535–585.

Vosniadou, S., Ioannides, C., Dimitrakopoulou, A., & Papademetriou, E.(2001). Designing learning environments to promote conceptual changein science. Learning and Instruction, 15, 317–419.

Vygotsky, L. (1978). Mind and society: The development of higher psy-chological processes. Cambridge, MA: Harvard University Press.

Wandersee, J., Mintzes, J., & Novak, J. (1994). Research on alterna-tive conceptions in science. In D. Gabel (Ed.), Handbook of researchon science teaching and learning (pp. 177–210). New York: Mac-millan.

Wood, D., Bruner, J., & Ross, G. (1976). The role of tutoring in problemsolving. Journal of Child Psychology and Psychiatry and Allied Disci-plines, 17, 89–100.

(Appendix follows)

325EFFECTS OF INSTRUCTIONAL SUPPORT

Page 20: Effects of Instructional Support Within Constructivist Learning … · 2009-03-18 · Effects of Instructional Support Within Constructivist Learning Environments for Elementary School

Appendix

Coding Scheme for Teacher (T) and Student (S) Statements During Class Discussions

Source Type Classification Examples

Teacher Structuring Comments on observations, descriptions ofexperiments, or requests to repeat results

What happened exactly?What did you feel when you immersed something into water?Can you please demonstrate this?

Requests explanations and justifications How come that the water rises?Why does the water push against your hand?

Points to transfer or application Have you observed this also in a different experiment?Where else did you notice this?

Fosters reasoning processes of generalization Is this correct: The water pushes all objects upwards?You said that the water raises with a stone, with a piece of clay, with a pot,

with a ball; does the water always rise?How else could you express this: the larger an object, the more . . .?

Addresses misconceptions by empiricalconfrontation or by calling attention toconflicting results

You said that all light things will float (immerses a metal pin into water).What about a ship that rests in the harbor?

Underlines important comments What Kathrin just said was really important.Yes, the water does push against it.

Focuses attention on specific aspects Look really closely at what happens when I immerse this pot.What happens to the rubber band? Look closely.

Elaborates on concepts You said that the water is pushed up or that the water is pushed away. Wealso say that the water is displaced.

Helps with other means (drawing, gesture,demonstration)

Show it with your hands.This is important; we will talk about this in a minute; now first finish your

thought.Why don’t you show it with arrows on the board?

Reinforces student comments Well observed.Good idea.

Passive Poses an open question Did anyone find out something new?Who would like to report what you have observed?

Turns over question to another student/students Max said it may be because of the air. What do the others think about this?Checks comprehension You said it’s because of its form; what do you mean by this?Repeats You mean that it is because of the size, not the weight (of the object).Requests repetition Could you say this again?

Could you show this again so that everyone can see?Encourages to rethink Who has got another idea? All of your ideas are important.Responds to student question S. Are they all the same size?

T: Yes, they are all the same size.

Managerial Discipline Please be quiet.Now listen and stop the noise.

Task You are supposed to weigh the cubes and sort them according to their weight.Organizational We’d best take this container and mark it.

Please put your folders under your seats.

Student Active Offers own ideas How come that this piece of wood is lighter than the other one?How can this be that metal will sink but a ship floats? It’s made of iron, too.I discovered something; this container displaces the same amount of water that

fits into it.Answers open teacher question T: Who would like to share their ideas?

S: I made a ship of clay and then I tried out whether it will sink in water withdish soap.

Reaction tostudent

Responds to student comment with ownideas

S1: I think it’s because of the salty water.S2: No, that can’t be because ships will also float in rivers.

Reinforces student comments What Teresa said about the space I think is really good.Reaction to

teacherT: What did you find out?S: I found out that the water rises.

Other Organizational I need this container and a pen.Observational He is loading his boat.

This doesn’t fit into the container.Other Be quiet.

Received February 1, 2005Revision received January 31, 2006

Accepted February 2, 2006 �

326 HARDY, JONEN, MOLLER, AND STERN